9 |
use strict; |
use strict; |
10 |
use Data::Dumper; |
use Data::Dumper; |
11 |
use Carp; |
use Carp; |
12 |
|
use Getopt::Long; |
13 |
|
|
14 |
use lib './lib'; |
use lib './lib'; |
15 |
use WebPAC; |
use WebPAC; |
16 |
|
use WebPAC::jsFind; |
17 |
|
use WebPAC::Index; |
18 |
|
use WebPAC::Tree; |
19 |
|
|
20 |
|
# options which can be changed via command line |
21 |
|
# |
22 |
|
my $code_page = 'ISO-8859-2'; |
23 |
|
my ($limit_mfn, $start_mfn, $debug, $low_mem); |
24 |
|
my $index_path = './out/index'; |
25 |
|
|
26 |
|
my $result = GetOptions( |
27 |
|
"code_page=s" => \$code_page, |
28 |
|
"limit_mfn=i" => \$limit_mfn, |
29 |
|
"start_mfn=i" => \$start_mfn, |
30 |
|
"debug!" => \$debug, |
31 |
|
"low_mem!" => \$low_mem, |
32 |
|
); |
33 |
|
|
34 |
|
# create WebPAC object |
35 |
|
# |
36 |
my $webpac = new WebPAC( |
my $webpac = new WebPAC( |
37 |
code_page => 'ISO-8859-2', |
code_page => $code_page, |
38 |
limit_mfn => 500, |
limit_mfn => $limit_mfn, |
39 |
# debug => 1, |
start_mfn => $start_mfn, |
40 |
|
debug => $debug, |
41 |
|
low_mem => $low_mem, |
42 |
|
) || die; |
43 |
|
|
44 |
|
my $log = $webpac->_get_logger() || die "can't get logger"; |
45 |
|
|
46 |
|
$log->debug("creating WebPAC::jsFind object"); |
47 |
|
|
48 |
|
my $index = new WebPAC::jsFind( |
49 |
|
index_path => $index_path, |
50 |
|
keys => 10, |
51 |
) || die; |
) || die; |
52 |
|
|
53 |
my $log = $webpac->_get_logger(); |
my $thes; |
54 |
|
|
55 |
$|=1; |
$|=1; |
56 |
|
|
62 |
# { 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
# { 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
63 |
# { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
# { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
64 |
{ 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
{ 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
65 |
|
{ 'key' => '900_mfn:v900', 'val' => 'v000' }, |
66 |
|
# tree structure |
67 |
|
{ 'eval' => 'length("v251") == 2', 'key' => 'root:v251', 'val' => 'v900' }, |
68 |
|
{ 'eval' => '"v251"', 'key' => 'code:v900', 'val' => 'v561^4:v251' }, |
69 |
], |
], |
70 |
); |
); |
71 |
|
|
72 |
|
$log->debug("isis file ",$webpac->{'isis_filename'}," opened"); |
73 |
|
|
74 |
$log->info("rows: $maxmfn"); |
$log->info("rows: $maxmfn"); |
75 |
|
|
76 |
$webpac->open_import_xml(type => 'isis_hidra_ths'); |
$webpac->open_import_xml(type => 'isis_hidra_ths'); |
77 |
|
|
78 |
|
if(1) { # XXX |
79 |
|
|
80 |
while (my $rec = $webpac->fetch_rec) { |
while (my $rec = $webpac->fetch_rec) { |
81 |
|
|
82 |
my @ds = $webpac->data_structure($rec); |
my @ds = $webpac->data_structure($rec); |
83 |
|
|
84 |
if ($log->is_debug) { |
if (0 && $log->is_debug) { |
85 |
$log->debug("rec = ",Dumper($rec)); |
$log->debug("rec = ",Dumper($rec)); |
86 |
$log->debug("ds = ",Dumper(\@ds)); |
$log->debug("ds = ",Dumper(\@ds)); |
87 |
} |
} |
88 |
|
|
89 |
print $webpac->output( |
next if (! @ds); |
90 |
template => 'text.tt', |
|
91 |
data => \@ds, |
my $filename = $webpac->{'current_filename'} || $log->logdie("no current_filename in webpac object"); |
92 |
) if (@ds); |
|
93 |
|
if ($filename) { |
94 |
|
$webpac->output_file( |
95 |
|
file => $filename, |
96 |
|
template => 'html.tt', |
97 |
|
data => \@ds, |
98 |
|
headline => $webpac->{'headline'}, |
99 |
|
); |
100 |
|
} else { |
101 |
|
print $webpac->output( |
102 |
|
template => 'text.tt', |
103 |
|
data => \@ds, |
104 |
|
headline => $webpac->{'headline'}, |
105 |
|
); |
106 |
|
} |
107 |
|
|
108 |
|
my $headline = $webpac->{'headline'}; |
109 |
|
|
110 |
|
my $f = $filename; |
111 |
|
$f =~ s!out/!!; |
112 |
|
|
113 |
|
# save into index |
114 |
|
foreach my $ds (@ds) { |
115 |
|
next if (! $ds->{'swish'}); |
116 |
|
|
117 |
|
$index->insert( |
118 |
|
index_name => $ds->{'tag'}, |
119 |
|
path => $f, |
120 |
|
headline => $headline, |
121 |
|
words => join(" ",@{$ds->{'swish'}}) |
122 |
|
); |
123 |
|
} |
124 |
|
|
125 |
|
# save into sorted index (thesaurus) |
126 |
|
foreach my $ds (@ds) { |
127 |
|
next if (! $ds->{'index'}); |
128 |
|
|
129 |
|
$thes->{$ds->{'tag'}} ||= new WebPAC::Index; |
130 |
|
|
131 |
|
foreach my $h (@{$ds->{'index'}}) { |
132 |
|
$thes->{$ds->{'tag'}}->insert( |
133 |
|
path => $f, |
134 |
|
headline => $h, |
135 |
|
); |
136 |
|
} |
137 |
|
} |
138 |
|
|
139 |
|
# print Dumper(\@ds); |
140 |
|
|
141 |
|
} |
142 |
|
|
143 |
|
foreach my $t (keys %{$thes}) { |
144 |
|
|
145 |
|
my @e = $thes->{$t}->elements; |
146 |
|
if (! @e) { |
147 |
|
$log->logwarn("no elements in sorted index $t?"); |
148 |
|
next; |
149 |
|
} |
150 |
|
|
151 |
|
my $file = "./out/bfilter/$t.txt"; |
152 |
|
$log->info("saving sorted index $t to '$file' [".scalar(@e)." elements]"); |
153 |
|
|
154 |
|
$webpac->output_file( |
155 |
|
file => $file, |
156 |
|
template => 'index.tt', |
157 |
|
data => \@e, |
158 |
|
index_name => $t, |
159 |
|
); |
160 |
} |
} |
161 |
|
|
162 |
if ($log->is_debug) { |
if (0 && $log->is_debug) { |
163 |
$log->debug("lookup hash: ",Dumper($webpac->{'lookup'})); |
$log->debug("lookup hash: ",Dumper($webpac->{'lookup'})); |
164 |
$log->debug("data hash: ",Dumper($webpac->{'data'})); |
$log->debug("data hash: ",Dumper($webpac->{'data'})); |
165 |
|
foreach my $t (keys %{$thes}) { |
166 |
|
$log->debug("thesaurus $t hash: ",Dumper($thes->{$t})); |
167 |
|
} |
168 |
} |
} |
169 |
|
|
170 |
|
} # XXX if(0) |
171 |
|
|
172 |
|
$log->debug("lookup hash: ",Dumper($webpac->{'lookup'})); |
173 |
|
|
174 |
|
$log->info("creating tree"); |
175 |
|
|
176 |
|
# |
177 |
|
# define tree structure |
178 |
|
# |
179 |
|
|
180 |
|
my $l = $webpac->{'lookup'} || $log->logconfess("can't find lookup"); |
181 |
|
|
182 |
|
my @tree = ({ |
183 |
|
# level 0 |
184 |
|
code_arr => sub { sort keys %{$l} }, |
185 |
|
filter_code => sub { |
186 |
|
my $t = shift; |
187 |
|
return $t if ($t =~ s/root://); |
188 |
|
}, |
189 |
|
lookup_v900 => sub { shift @{$l->{"root:".$_[0]}} }, |
190 |
|
lookup_term => sub { shift @{$l->{"d:".$_[1]}} }, |
191 |
|
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[1]}} }, |
192 |
|
have_children => sub { defined($l->{"a:".$_[0]."::"}) }, |
193 |
|
child_code => sub { return $_[0] }, |
194 |
|
style => 'display: none', |
195 |
|
},{ |
196 |
|
# 1 |
197 |
|
code_arr => sub { @{$l->{"a:".$_[0]."::"}} }, |
198 |
|
filter_code => sub { shift }, # nop |
199 |
|
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
200 |
|
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
201 |
|
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
202 |
|
have_children => sub { defined($l->{"a:".$_[1].":"}) }, |
203 |
|
child_code => sub { return $_[1] }, |
204 |
|
style => 'display: none', |
205 |
|
},{ |
206 |
|
# 2 |
207 |
|
code_arr => sub { @{$l->{"a:".$_[0].":"}} }, |
208 |
|
filter_code => sub { shift }, |
209 |
|
lookup_v900 => sub { shift }, |
210 |
|
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
211 |
|
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
212 |
|
have_children => sub { defined($l->{"a:".$_[2].":".$_[1]}) }, |
213 |
|
child_code => sub { return "a:".$_[2].":".$_[1] }, |
214 |
|
style => 'display: none', |
215 |
|
},{ |
216 |
|
# 3 uži pojam |
217 |
|
code_arr => sub { @{$l->{$_[0]}} }, |
218 |
|
filter_code => sub { shift }, |
219 |
|
lookup_v900 => sub { |
220 |
|
my ($c,$p) = @_; |
221 |
|
$p =~ s/^a:(..:....):.*$/$1/; |
222 |
|
return "a:".$p.":".$c; |
223 |
|
}, |
224 |
|
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
225 |
|
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
226 |
|
have_children => sub { defined($l->{$_[1]}) }, |
227 |
|
child_code => sub { return $_[1] }, |
228 |
|
},{ |
229 |
|
# 4 |
230 |
|
code_arr => sub { @{$l->{$_[0]}} }, |
231 |
|
filter_code => sub { shift }, |
232 |
|
lookup_v900 => sub { |
233 |
|
my ($c,$p) = @_; |
234 |
|
$p =~ s/^a:(..:....):.*$/$1/; |
235 |
|
return "a:".$p.":".$c; |
236 |
|
}, |
237 |
|
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
238 |
|
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
239 |
|
have_children => sub { defined($l->{$_[1]}) }, |
240 |
|
child_code => sub { return $_[1] }, |
241 |
|
},{ |
242 |
|
# 5 |
243 |
|
code_arr => sub { @{$l->{$_[0]}} }, |
244 |
|
filter_code => sub { shift }, |
245 |
|
lookup_v900 => sub { |
246 |
|
my ($c,$p) = @_; |
247 |
|
$p =~ s/^a:(..:....):.*$/$1/; |
248 |
|
return "a:".$p.":".$c; |
249 |
|
}, |
250 |
|
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
251 |
|
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
252 |
|
have_children => sub { defined($l->{$_[1]}) }, |
253 |
|
child_code => sub { return $_[1] }, |
254 |
|
},{ |
255 |
|
# 6 |
256 |
|
code_arr => sub { @{$l->{$_[0]}} }, |
257 |
|
filter_code => sub { shift }, |
258 |
|
lookup_v900 => sub { |
259 |
|
my ($c,$p) = @_; |
260 |
|
$p =~ s/^a:(..:....):.*$/$1/; |
261 |
|
return "a:".$p.":".$c; |
262 |
|
}, |
263 |
|
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
264 |
|
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
265 |
|
have_children => sub { defined($l->{$_[1]}) }, |
266 |
|
child_code => sub { return $_[1] }, |
267 |
|
},{ |
268 |
|
# 7 |
269 |
|
code_arr => sub { @{$l->{$_[0]}} }, |
270 |
|
filter_code => sub { shift }, |
271 |
|
lookup_v900 => sub { |
272 |
|
my ($c,$p) = @_; |
273 |
|
$p =~ s/^a:(..:....):.*$/$1/; |
274 |
|
return "a:".$p.":".$c; |
275 |
|
}, |
276 |
|
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
277 |
|
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
278 |
|
have_children => sub { defined($l->{$_[1]}) }, |
279 |
|
child_code => sub { return $_[1] }, |
280 |
|
},{ |
281 |
|
# 8 |
282 |
|
code_arr => sub { @{$l->{$_[0]}} }, |
283 |
|
filter_code => sub { shift }, |
284 |
|
lookup_v900 => sub { |
285 |
|
my ($c,$p) = @_; |
286 |
|
$p =~ s/^a:(..:....):.*$/$1/; |
287 |
|
return "a:".$p.":".$c; |
288 |
|
}, |
289 |
|
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
290 |
|
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
291 |
|
# have_children => sub { defined($l->{$_[1]}) }, |
292 |
|
# child_code => sub { return $_[1] }, |
293 |
|
have_children => sub { 0 }, |
294 |
|
child_code => sub { 0 }, |
295 |
|
}); |
296 |
|
|
297 |
|
my $tree = new WebPAC::Tree( |
298 |
|
dir => './out', |
299 |
|
html => 'browse.html', |
300 |
|
template => './output_template/tree.tt', |
301 |
|
js => 'tree-ids.js', |
302 |
|
tree => \@tree, |
303 |
|
); |
304 |
|
|
305 |
|
$log->info("closing index"); |
306 |
|
$index->close; |
307 |
|
|
308 |
|
$log->info("elapsed time: ",$webpac->fmt_time(time() - $webpac->{'start_t'})); |